Multiple community-based approaches can aid in quantifying mortality in the absence of reliable health facility data. Community-based sentinel site surveillance that was used to document mortality and the systems utility for outbreak detection was evaluated. We retrospectively analyzed data from 46 sentinel sites in three sous-préfectures with a reinforced malaria control program and one sous-préfecture without (Koundou) in Guinea. Deaths were recorded by key informants and classified as due to malaria or another cause. Malaria deaths were those reported as due to malaria or fever in the 3 days before death with no other known cause. Suspect Ebola virus disease (sEVD) deaths were those due to select symptoms in the EVD case definition. Deaths were aggregated by sous-préfecture and analyzed by a 6-month period. A total of 43,000 individuals were monitored by the surveillance system; 1,242 deaths were reported from July 2011-June 2014, of which 55.2% (N = 686) were reported as due to malaria. Malaria-attributable proportional mortality decreased by 26.5% (95% confidence interval [CI] = 13.9-33.1, P < 0.001) in the program area and by 6.6% (95% CI = -17.3-30.5, P = 0.589) in Koundou. Sixty-eight deaths were classified as sEVD and increased by 6.1% (95% CI = 1.3-10.8, P = 0.021). Seventeen sEVD deaths were reported from November 2013 to March 2014 including the first two laboratory-confirmed EVD deaths. Community surveillance can capture information on mortality in areas where data collection is weak, but determining causes of death remains challenging. It can also be useful for outbreak detection if timeliness of data collection and reporting facilitate real-time data analysis.